Why Measuring Poverty Isn’t That Simple

Although it seems intuitive, defining the concept of poverty has posed economists difficulty. With a great deal of research in past decades, the measurement of poverty has evolved greatly over time. Although many methods of the past are still in use today, new techniques have been developed to provide insights which simpler measures never could.

Perhaps the simplest way to define poverty would be to set an income poverty line and declare anyone whose income is below that line to be in poverty. This strict and absolute cut off would suggest poverty occurs when one is deprived of a fixed amount. This fixed amount  would be determined by the income required to not only survive but also participate in society. This suggests that each country would need a different income poverty line, as being able to purchase a computer is arguably socially necessary to live in a country like Australia, but socially unnecessary in other countries. However, this would render this measure of poverty useless in helping economists to compare poverty levels across different countries.

This brings us to the idea that a measure of poverty should account not only for absolute but also relative poverty.  A relative measurement of poverty would be defined relative to some population benchmark. This could be, for example, earning income some percentage below the median. (1) However, approaching the measurement of poverty from a purely relative perspective is also insufficient. If the incomes of the whole population were suddenly multiplied by a thousandfold, such a relative measure would remain unchanged, even though everyone in the country had grown richer. Consequently, although relative measures of poverty can account for inequality, a successful method still needs to incorporate the absolute aspect of poverty to account for economic growth.

Relative measures of poverty can also fail to account for the intensity of poverty. (2) A primary example is the headcount ratio (HCR), which defines a poverty line p and calculates the percentage of a country’s population with a mean income below that line. Because the measure is country specific, each country’s p can be tailored to reflect each country’s price of a basket of necessities. The HCR therefore seems to satisfy the requirements for a relative measure of poverty, allowing for comparisons of poverty on a global scale. The data required for the HCR method is also easily available, and it has become a highly prevalent tool in poverty analysis.

However, the HCR fails to distinguish between individuals below the poverty line and consequently ignores the varying degrees of poverty faced by different demographics. Renowned development economist Amartya Sen noted that a government utilising only the HCR ‘faces a strong temptation to concentrate on the richest among the poor, since this is the way that the number of the poor… can be most easily reduced’. (3) It would be simpler for a government to slightly bolster the incomes of individuals who were only slightly under the poverty line p and greatly reduce poverty as measured by HCR. Why bother boosting the incomes of those who are most impoverished when it won’t be reflected as improvements in the HCR? Clearly, a more adequate measure of poverty must also consider the intensity of poverty.

The poverty gap ratio (PGR) and income gap ratio (IGR) are two methods which capture poverty intensity. Both methods utilise a poverty line p, similar to the HCR. The PGR considers the average distance of an impoverished individual’s income from p and expresses it as a percentage of p. A higher PGR indicates a greater depth of poverty. (4) IGR takes a similar approach but expresses the average income shortfall as a percentage of the total income earned by all impoverished individuals if they were lifted out of poverty. Because it gives the total amount of income needed to remove the presence of poverty, IGR is generally regarded as more accurate. However, even these measures distort the varying levels of poverty. A small population of individuals all similarly far below the poverty line could return similar results to a large impoverished population with greater income variance.

In 1976, Sen proposed an axiomatic approach to poverty measurement. (5) These axioms accounted for the different levels of poverty in a way the previous ratio-based measures did not. Although not all of them are discussed here, the following axioms highlight some of the flaws in aforementioned measures. The focus axiom states that any poverty measurement must be independent of those above the poverty line. This is a logical requirement of poverty analysis, which is passed by all three ratio-based methods. The monotonicity axiom states that an increase in income below the poverty line must reduce poverty. Both the PGR and IGR satisfy this axiom while the HCR does not. If a poor person received an increase in income but was not lifted above the poverty line, the HCR would not change. The transfer axiom states that a transfer of income from someone below the poverty line to anyone who is richer will result in an increase in poverty. A transfer of income from a poor individual to a wealthier person below the poverty line would not affect measurements taken by the HCR, PGR, or IGR methods. As such, they all violate the transfer axiom. The strength in these axioms is that they measure not just the absolute assets of the poor, but the often-overlooked inequality between them. (6)

Since Sen’s axioms, many new ways to measure poverty have been proposed. (7) These include the FGT index, which gives a stronger weighting to the most impoverished individuals, and the SST index, which combines IGR techniques with Gini coefficient measurements. Also significant is the relatively recent Global Multidimensional Poverty Index, which allows economists to weight different dimensions of poverty such lack of education or healthcare differently. Although the weights can be viewed as arbitrary as they are subject to the values of the policymakers who determine them, the GMPI can capture a more nuanced picture of poverty than the other measures.  Although no measure satisfies every established axiom, these axioms have given rise to measures which better account for both relative and absolute poverty.

Photo by Jordan Opel on Unsplash

(1) ReferencesFoster, J.E. (1998). Absolute versus Relative Poverty. The American Economic Review, [online] 88(2). Available at: https://www.jstor.org/stable/116944?seq=3#metadata_info_tab_contents.

(2) mdgs.un.org. (2013). 1.2 Poverty gap ratio. [online] Available at: http://mdgs.un.org/unsd/mi/wiki/1-2-Poverty-gap-ratio.ashx?From=Indicator-1-2-poverty-gap-ratio [Accessed 18 Aug. 2020].

(3) Bellù, L.G. and Liberati, P. (2005). Impacts of Policies on Poverty. EASYPol, [online] pp.1–8. Available at: http://www.fao.org/docs/up/easypol/321/axioms_pov_msmt_008en.pdf

(4) Ray, D. (1998). Development economics. Princeton, N.J.: Princeton University Press.‌

(5) Xu, K. and Osberg, L. (2002). On Sen’s Approach to Poverty Measures and Recent Developments. China Economic Quarterly, 1.

(6) Xu, K. (1998). Statistical inference for the Sen-Shorrocks-Thon index of poverty intensity. Journal of Income Distribution, [online] 8(1), pp.143–152. Available at: https://www.mathstat.dal.ca/~kuan/SSTIndex.pdf [Accessed 17 Aug. 2020].

(7) Sen, A. (1976). Poverty: An Ordinal Approach to Measurement. Econometrica, 44(2).‌